-
Notifications
You must be signed in to change notification settings - Fork 0
/
PupilDetector.py
181 lines (148 loc) · 5.87 KB
/
PupilDetector.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
import dlib
import cv2
import numpy as np
import time
from PupilDetector2 import GradientIntersect
RIGHT_EYE = list(range(36, 42)) # 6
LEFT_EYE = list(range(42, 48)) # 6
def shape_to_np(shape, dtype="int"):
# initialize the list of (x, y)-coordinates
coords = np.zeros((shape.num_parts, 2), dtype=dtype)
# loop over all facial landmarks and convert them
# to a 2-tuple of (x, y)-coordinates
for i in range(0, shape.num_parts):
coords[i] = (shape.part(i).x, shape.part(i).y)
# return the list of (x, y)-coordinates
return coords
detector = dlib.get_frontal_face_detector()
predictor = dlib.shape_predictor("./dlib/shape_predictor_68_face_landmarks.dat")
cap = cv2.VideoCapture(0)
time.sleep(1) # warming up
if not cap.isOpened():
exit()
ttimecount = 0
FRAME_REPEAT = 30
tfps = 30
available = 0
viewType = 0
# loc = (0,0)
eyeConst = 3
loc_l = (0,0)
loc_r = (0,0)
gi_l = GradientIntersect()
gi_r = GradientIntersect()
while True:
if(ttimecount == 0):
starttime = time.time()
ttimecount += 1
cap.retrieve()
ret, image = cap.read()
# if(ttimecount%2 == 0):
# continue
# image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
if not ret:
break
# tempWidth = 80
# available = objEyeTrack.preprocess(img, (160-tempWidth,120-tempWidth,320+tempWidth,240+tempWidth))
# available = objEyeTrack.preprocess(img, (0,0, 640 , 480))
# if(available > 0 ):
# objEyeTrack.temp_run(image, img, gazeType=viewType )
# objEyeTrack.randering(image)
faces = detector(gray)
for face in faces:
p_lefteye = []
p_righteye = []
x1, y1, x2, y2 = face.left(), face.top(), face.right(), face.bottom()
cv2.rectangle(image, (x1, y1), (x2, y2), (0, 0, 255), thickness=2)
landmarks = predictor(gray, face)
tlandmark = shape_to_np(landmarks)
# for n in range(0, 68):
# x = landmarks.part(n).x
# y = landmarks.part(n).y
# cv2.circle(image, (x, y), 4, (255, 0, 0), -1)
for (sX, sY) in tlandmark:
cv2.circle(image, (sX, sY), 1, (255, 0, 0), -1)
for n in LEFT_EYE:
x = landmarks.part(n).x
y = landmarks.part(n).y
p_lefteye.append([x,y])
for n in RIGHT_EYE:
x = landmarks.part(n).x
y = landmarks.part(n).y
p_righteye.append([x,y])
p_lefteye = np.array(p_lefteye)
p_righteye = np.array(p_righteye)
# print(p_lefteye, p_lefteye.shape)
# p_lefteye.append([face[t] for t in LEFT_EYE])
# p_righteye.append([face[t] for t in RIGHT_EYE])
# for tpoint in p_lefteye:
lx = np.min(p_lefteye, axis=0)[0] - eyeConst
ly = np.min(p_lefteye, axis=0)[1] - eyeConst # lip_check
lx2 = np.max(p_lefteye, axis=0)[0] + eyeConst
ly2 = np.max(p_lefteye, axis=0)[1] + eyeConst
print(lx, ly, lx2, ly2)
# leye_result = eye_aspect_ratio(tpoint)
# p_lefteye_local = (p_lefteye - np.array([x, y]))
# print('crop_lefteye', p_lefteye_local)
clipping_gray_l = gray[ly:ly2, lx:lx2]
# if (SAVE_PART_OF_EYES):
# cv2.imwrite('eyeL{:02d}.png'.format(tnum), clipping_gray)
rx = np.min(p_righteye, axis=0)[0] - eyeConst
ry = np.min(p_righteye, axis=0)[1] - eyeConst # lip_check
rx2 = np.max(p_righteye, axis=0)[0] + eyeConst
ry2 = np.max(p_righteye, axis=0)[1] + eyeConst
# print(lx, ly, lx2, ly2)
clipping_gray_r = gray[ry:ry2, rx:rx2]
if(available == 0):
if(loc_l[0]==0 and loc_l[1]==0):
# gi_l = GradientIntersect()
loc_l = gi_l.locate(clipping_gray_l)
print('loc_l', loc_l)
cv2.circle(image, (int(lx + loc_l[1]), int(ly + loc_l[0])), 2, (0, 0, 255), -1)
available = 0
else:
# loc_l = gi_l.track(clipping_gray_l, loc_l)
# cv2.circle(image, (int(loc_l[1]+lx), int(loc_l[0]+ly)), 2, (0, 255, 255), -1)
loc_l2 = gi_l.track(gray, (ly+loc_l[0], lx+loc_l[1]))
cv2.circle(image, (int(loc_l2[1]), int(loc_l2[0])), 2, (0, 255, 255), -1)
# loc_l2 = gi_l.track(gray[y1:y2, x1:x2], (ly +loc_l[0]-y1, lx +loc_l[1]-x1))
# cv2.circle(image, (int(x1+lx+loc_l[1]), int(y1+ly+loc_l[0])), 2, (0, 255, 255), -1)
if(loc_r[0]==0 and loc_r[1]==0):
# gi_r = GradientIntersect()
loc_r = gi_r.locate(clipping_gray_r)
print('loc_r', loc_r)
cv2.circle(image, (int(rx + loc_r[1]), int(ry + loc_r[0])), 2, (0, 0, 255), -1)
available = 0
else:
# loc_r = gi_r.track(clipping_gray_r, loc_r)
loc_r2 = gi_r.track(gray, (ry+loc_r[0] , rx+loc_r[1]))
cv2.circle(image, (int(loc_r2[1]), int(loc_r2[0])), 2, (0, 255, 255), -1)
# cv2.imshow("result", clipping_gray)
# cv2.waitKey(0)
# elif(available == 1):
# loc_l = gi_l.track(gray, (int(lx + loc_l[1]), int(ly + loc_l[0])))
# print('loc2',loc_l)
# image = cv2.cvtColor(img, cv2.COLOR_GRAY2RGB)
cv2.putText(image, 'FPS={:.1f} {:s}'.format(tfps, " "),
(10, 460),
cv2.FONT_HERSHEY_COMPLEX_SMALL, 1, (0, 255, 0), thickness=2, lineType=8)
if (ttimecount >= FRAME_REPEAT):
tfps = ttimecount / (time.time() - starttime)
ttimecount = 0
cv2.imshow('image', image)
# gi = GradientIntersect()
# loc = gi.locate(gray)
# print(loc)
#
# ret, frame = cap.read()
# gray = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)
# cv2.imshow('image', frame)
#
# loc = gi.track(gray, loc)
# print(loc)
key = cv2.waitKey(1)
if key == ord('q'):
break
cap.release()
cv2.destroyAllWindows()